ad-research
Pass
Audited by Gen Agent Trust Hub on Sep 21, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted text and transcripts from external ad libraries and landing pages, creating a vulnerability surface where malicious instructions could be embedded in competitor content.
- Ingestion points: Scrapes ad copy, headlines, and video transcripts from Meta Ad Library, TikTok, and Google search results (SKILL.md, Steps 2 and 3).
- Boundary markers: The prompt instructions do not provide explicit delimiters or "ignore embedded instructions" warnings for the agent when processing raw scraped text.
- Capability inventory: The skill uses
playwright MCPfor web navigation,memory MCPfor long-term pattern storage, andsequential MCPfor synthesis, allowing scraped content to influence stored knowledge and downstream reports. - Sanitization: There is no evidence of filtering or sanitizing the extracted transcripts or body copy before they are processed by the reasoning agents.
Audit Metadata